LEARNING-CIRCLE-24 PROJECT

Learning-Circle-24

Contributors

Crime in Chicago¶

This is a dataset The data dictionary can be found in Chicago Crime Portal

In this notebook, we are going to explore more about crime in Chicago and try to answer few questions.

Analyzing the Chicago crime dataset involves examining various aspects of the data to understand patterns, trends, and characteristics of criminal activities in Chicago. We perform analysis using Python, pandas and visualization libraries using the following steps:

  1. Load the Dataset: Load the Chicago crime dataset into a DataFrame.
  2. Explore the Dataset: Examine the structure, columns, and basic statistics of the dataset.
  3. Data Preprocessing: We cleaned the data by dropping redundant columns,handling missing values and converting columns to their appropriate format.
  4. Feature Engineering: Create new column for Months, Day, Season
  5. EDA: Exploratory Data Analysis(Non-Graphical)
    • What is the total number of crime cases registered by the police department from 2001 to 2022?
    • What is the number of unique crime cases?
    • What is the number of non-unique cases?
    • Out of total case registered how many are arrested and how many are not?
    • How many unique crime type is available and what is the count for each crime from 2001-2022?
    • what is the number of crime per year?
    • What is the number of crime per month?
    • what is the number of crime by season?
    • what is the number of crime per day?
    • What are the top 10 most common crime types?
    • what are the top 10 most common location description for crimes?
    • what are the top 10 most common crimes in 2021?
    • How many crimes were reported each month in 2021 ?
    • what is the top 10 days with the most crime count ?
    • what are the Crimes with the most Arrest?
    • Count of Arrest and Non arrest for each Crime type.
    • Percentage of Arrest and Non arrest for each Crime type.
  • Exploratory Data Analysis(Graphical)
    • Trend of crimes in chicago from 2001 to 2022(Line chart)
    • Monthly crime trend(Line chart)
    • Seasonal crime trend(Line chart)
    • Daily Crime trend(Line chart)
    • Top 10 Days with the Most Crime Counts Each Day(Bar chart)
    • Top 6 Most reported Crime(ggplot)
    • Top 10 common crime by type(Pie Chart)
    • Top 10 Common location for crimes
      1. Top 10 Location Description for crimes
      2. Top 10 Community Area for crimes
      3. Top 10 Ward for crimes
      4. Top 10 Block for crimes
    • Top - 10 communities with the most crime in 2021
    • Top - 5 Primary Crime Types and Their Locations in Chicago (2021)
    • Top - 10 Location where theft normally occurs
    • Top - 10 Location for theft in Chicago in 2021
    • Number of crimes per Year x Months
    • Are there specific location that crimes used to happen?
    • Count of Arrest and Non-Arrest for Each Crime Type
  1. Summarize and Visualize Insights: Summarize and Visualize the key insights and findings from the analysis using plots and charts.
    • How has crime in Chicago changed across years? Was 2021 really the bloodiest year in two decades?
    • Are some types of crimes more likely to happen in specific locations or specific time of the day or specific day of the week, or specific month of the year than other types of crimes?
In [1]:
# importing required libraries
import pandas as pd
import numpy as np
import seaborn as sns
import plotly.express as px
import plotly.graph_objects as go
from plotly.subplots import make_subplots
import matplotlib.pyplot as plt
%matplotlib inline
from datetime import datetime
In [2]:
# Set display option
pd.set_option('display.max.columns', None)

1. Loading the dataset¶

This dataset was extracted in 2023 and contains crime cases from 2001 to 2023, but we want the data for only cases from 2001 to 2022 due to incomplete cases for 2023 in order to avoid a bias analysis.

In [3]:
full_dataset = pd.read_csv('../Data-Science-Projects/dataset/crime_data_chicago.csv')
# Drop rows that have 2023 as the year.
chicago_crime_df = pd.DataFrame(full_dataset[full_dataset['Year'] != 2023])
chicago_crime_df.head()
Out[3]:
Unnamed: 0 ID Case Number Date Block IUCR Primary Type Description Location Description Arrest Domestic Beat District Ward Community Area FBI Code X Coordinate Y Coordinate Year Updated On Latitude Longitude Location
0 0 6407111 HP485721 07/26/2008 02:30:00 PM 085XX S MUSKEGON AVE 1320 CRIMINAL DAMAGE TO VEHICLE STREET False False 423 4.0 10.0 46.0 14 1196638.0 1848800.0 2008 02/28/2018 03:56:25 PM 41.739980 -87.555120 (41.739979622, -87.555120042)
1 1 11398199 JB372830 07/31/2018 10:57:00 AM 092XX S ELLIS AVE 143C WEAPONS VIOLATION UNLAWFUL POSS AMMUNITION POOL ROOM True False 413 4.0 8.0 47.0 15 1184499.0 1843935.0 2018 08/07/2018 04:02:59 PM 41.726922 -87.599747 (41.726922145, -87.599746995)
2 2 5488785 HN308568 04/27/2007 10:30:00 AM 062XX N TRIPP AVE 0610 BURGLARY FORCIBLE ENTRY RESIDENCE True False 1711 17.0 39.0 12.0 05 1146911.0 1941022.0 2007 02/28/2018 03:56:25 PM 41.994138 -87.734959 (41.994137622, -87.734959049)
3 3 11389116 JB361368 07/23/2018 08:55:00 AM 0000X N KEELER AVE 0560 ASSAULT SIMPLE NURSING HOME/RETIREMENT HOME False False 1115 11.0 28.0 26.0 08A 1148388.0 1899882.0 2018 07/30/2018 03:52:24 PM 41.881217 -87.730590 (41.881217483, -87.730589961)
4 4 12420431 JE297624 07/11/2021 06:40:00 AM 016XX W HARRISON ST 051A ASSAULT AGGRAVATED - HANDGUN PARKING LOT / GARAGE (NON RESIDENTIAL) False False 1231 12.0 27.0 28.0 04A 1165430.0 1897441.0 2021 07/18/2021 04:56:02 PM 41.874174 -87.668082 (41.874173691, -87.668082118)

2. Explore the dataset¶

In [4]:
#checking the dimension of the dataset
chicago_crime_df.shape
Out[4]:
(2278726, 23)
In [5]:
# Display the first five rows of the dataset
chicago_crime_df.columns
Out[5]:
Index(['Unnamed: 0', 'ID', 'Case Number', 'Date', 'Block', 'IUCR',
       'Primary Type', 'Description', 'Location Description', 'Arrest',
       'Domestic', 'Beat', 'District', 'Ward', 'Community Area', 'FBI Code',
       'X Coordinate', 'Y Coordinate', 'Year', 'Updated On', 'Latitude',
       'Longitude', 'Location'],
      dtype='object')
In [6]:
# check the first five observations
chicago_crime_df.head()
Out[6]:
Unnamed: 0 ID Case Number Date Block IUCR Primary Type Description Location Description Arrest Domestic Beat District Ward Community Area FBI Code X Coordinate Y Coordinate Year Updated On Latitude Longitude Location
0 0 6407111 HP485721 07/26/2008 02:30:00 PM 085XX S MUSKEGON AVE 1320 CRIMINAL DAMAGE TO VEHICLE STREET False False 423 4.0 10.0 46.0 14 1196638.0 1848800.0 2008 02/28/2018 03:56:25 PM 41.739980 -87.555120 (41.739979622, -87.555120042)
1 1 11398199 JB372830 07/31/2018 10:57:00 AM 092XX S ELLIS AVE 143C WEAPONS VIOLATION UNLAWFUL POSS AMMUNITION POOL ROOM True False 413 4.0 8.0 47.0 15 1184499.0 1843935.0 2018 08/07/2018 04:02:59 PM 41.726922 -87.599747 (41.726922145, -87.599746995)
2 2 5488785 HN308568 04/27/2007 10:30:00 AM 062XX N TRIPP AVE 0610 BURGLARY FORCIBLE ENTRY RESIDENCE True False 1711 17.0 39.0 12.0 05 1146911.0 1941022.0 2007 02/28/2018 03:56:25 PM 41.994138 -87.734959 (41.994137622, -87.734959049)
3 3 11389116 JB361368 07/23/2018 08:55:00 AM 0000X N KEELER AVE 0560 ASSAULT SIMPLE NURSING HOME/RETIREMENT HOME False False 1115 11.0 28.0 26.0 08A 1148388.0 1899882.0 2018 07/30/2018 03:52:24 PM 41.881217 -87.730590 (41.881217483, -87.730589961)
4 4 12420431 JE297624 07/11/2021 06:40:00 AM 016XX W HARRISON ST 051A ASSAULT AGGRAVATED - HANDGUN PARKING LOT / GARAGE (NON RESIDENTIAL) False False 1231 12.0 27.0 28.0 04A 1165430.0 1897441.0 2021 07/18/2021 04:56:02 PM 41.874174 -87.668082 (41.874173691, -87.668082118)
In [7]:
 # Display column information and data types
chicago_crime_df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 2278726 entries, 0 to 2278725
Data columns (total 23 columns):
 #   Column                Dtype  
---  ------                -----  
 0   Unnamed: 0            int64  
 1   ID                    int64  
 2   Case Number           object 
 3   Date                  object 
 4   Block                 object 
 5   IUCR                  object 
 6   Primary Type          object 
 7   Description           object 
 8   Location Description  object 
 9   Arrest                bool   
 10  Domestic              bool   
 11  Beat                  int64  
 12  District              float64
 13  Ward                  float64
 14  Community Area        float64
 15  FBI Code              object 
 16  X Coordinate          float64
 17  Y Coordinate          float64
 18  Year                  int64  
 19  Updated On            object 
 20  Latitude              float64
 21  Longitude             float64
 22  Location              object 
dtypes: bool(2), float64(7), int64(4), object(10)
memory usage: 369.4+ MB
In [8]:
# checking missing value count
chicago_crime_df.isnull().sum()
Out[8]:
Unnamed: 0                   0
ID                           0
Case Number                  1
Date                         0
Block                        0
IUCR                         0
Primary Type                 0
Description                  0
Location Description      2877
Arrest                       0
Domestic                     0
Beat                         0
District                    12
Ward                    184695
Community Area          184267
FBI Code                     0
X Coordinate             23985
Y Coordinate             23985
Year                         0
Updated On                   0
Latitude                 23985
Longitude                23985
Location                 23985
dtype: int64
In [9]:
chicago_crime_df.describe()
Out[9]:
Unnamed: 0 ID Beat District Ward Community Area X Coordinate Y Coordinate Year Latitude Longitude
count 2.278726e+06 2.278726e+06 2.278726e+06 2.278714e+06 2.094031e+06 2.094459e+06 2.254741e+06 2.254741e+06 2.278726e+06 2.254741e+06 2.254741e+06
mean 1.139362e+06 6.882068e+06 1.186442e+03 1.129072e+01 2.272764e+01 3.752140e+01 1.164569e+06 1.885747e+06 2.009638e+03 4.184209e+01 -8.767161e+01
std 6.578117e+05 3.419168e+06 7.026836e+02 6.946692e+00 1.383464e+01 2.153282e+01 1.673955e+04 3.209855e+04 6.019724e+00 8.830434e-02 6.073538e-02
min 0.000000e+00 6.370000e+02 1.110000e+02 1.000000e+00 1.000000e+00 0.000000e+00 0.000000e+00 0.000000e+00 2.001000e+03 3.661945e+01 -9.168657e+01
25% 5.696812e+05 3.716076e+06 6.210000e+02 6.000000e+00 1.000000e+01 2.300000e+01 1.152948e+06 1.859053e+06 2.004000e+03 4.176866e+01 -8.771379e+01
50% 1.139362e+06 6.885990e+06 1.034000e+03 1.000000e+01 2.300000e+01 3.200000e+01 1.166060e+06 1.890673e+06 2.009000e+03 4.185578e+01 -8.766597e+01
75% 1.709044e+06 9.887568e+06 1.731000e+03 1.700000e+01 3.400000e+01 5.700000e+01 1.176365e+06 1.909219e+06 2.014000e+03 4.190668e+01 -8.762823e+01
max 2.278725e+06 1.278199e+07 2.535000e+03 3.100000e+01 5.000000e+01 7.700000e+01 1.205119e+06 1.951622e+06 2.022000e+03 4.202291e+01 -8.752453e+01

3. Data Preprocessing¶

Data Cleaning

Here are some steps that we would carry out in this phase

  1. Remove redundant or unnecessary columns
  2. Handle missing values
  3. Convert data types to the appropriate formats

Reasons for dropping the below columns

  • 'Unnamed: 0': Its just an index for individual row/ Observation
  • 'ID': Serial identifier for the record which is redundant due to the presence of a case number for each case.
  • 'Location': This is where the incident occurred in a format that allows for creation of maps and other geographic operation. It's the combination of both latitude and longitude feature of this dataset thus, not required.
  • 'Updated On': Holds the date and time the record was last updated.Which is not required for our analysis.
  • 'X Coordinate': The x coordinate is a shifted location from the actual location for partial redaction but falls on the same block. Since we have a longitude/latitude and the block where the crime happened, we can neglect this feature.
  • 'Y Coordinate': The y coordinate is a shifted location from the actual location for partial redaction but falls on the same block. Since we have a longitude/latitude and the block where the crime happened, we can neglect this feature.
  • 'IUCR': This is the Illinois Unifrom Crime Reporting code. This is directly linked to the Primary Type and Description, it provides information about the primary type and description of crime. Since primary type and description features already exist, it becomes a redundant column
  • Description: It's simply a subcategory of the primary type(which describes the type of crime committed). this can be omitted.
  • FBI Code: Removed because it is not necessary for our analysis as we are unable to de-code these codes.

1. Use a funtion to Drop redundant columns and/or columns that are not necessary for our analysis (a redundant column is one that gives infomation that another column already explains: eg ID and Case number)

In [10]:
chicago_crime_df.columns
Out[10]:
Index(['Unnamed: 0', 'ID', 'Case Number', 'Date', 'Block', 'IUCR',
       'Primary Type', 'Description', 'Location Description', 'Arrest',
       'Domestic', 'Beat', 'District', 'Ward', 'Community Area', 'FBI Code',
       'X Coordinate', 'Y Coordinate', 'Year', 'Updated On', 'Latitude',
       'Longitude', 'Location'],
      dtype='object')
In [11]:
# Dropping redundant columns
def drop_redundant_columns(df):
    redundant_columns = ['Unnamed: 0','ID', 'Location','Updated On', 'FBI Code', 'Description', 'X Coordinate','Y Coordinate','IUCR']
    df.drop(redundant_columns, axis=1, inplace= True)

drop_redundant_columns(chicago_crime_df)
In [12]:
chicago_crime_df.head()
Out[12]:
Case Number Date Block Primary Type Location Description Arrest Domestic Beat District Ward Community Area Year Latitude Longitude
0 HP485721 07/26/2008 02:30:00 PM 085XX S MUSKEGON AVE CRIMINAL DAMAGE STREET False False 423 4.0 10.0 46.0 2008 41.739980 -87.555120
1 JB372830 07/31/2018 10:57:00 AM 092XX S ELLIS AVE WEAPONS VIOLATION POOL ROOM True False 413 4.0 8.0 47.0 2018 41.726922 -87.599747
2 HN308568 04/27/2007 10:30:00 AM 062XX N TRIPP AVE BURGLARY RESIDENCE True False 1711 17.0 39.0 12.0 2007 41.994138 -87.734959
3 JB361368 07/23/2018 08:55:00 AM 0000X N KEELER AVE ASSAULT NURSING HOME/RETIREMENT HOME False False 1115 11.0 28.0 26.0 2018 41.881217 -87.730590
4 JE297624 07/11/2021 06:40:00 AM 016XX W HARRISON ST ASSAULT PARKING LOT / GARAGE (NON RESIDENTIAL) False False 1231 12.0 27.0 28.0 2021 41.874174 -87.668082
In [13]:
chicago_crime_df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 2278726 entries, 0 to 2278725
Data columns (total 14 columns):
 #   Column                Dtype  
---  ------                -----  
 0   Case Number           object 
 1   Date                  object 
 2   Block                 object 
 3   Primary Type          object 
 4   Location Description  object 
 5   Arrest                bool   
 6   Domestic              bool   
 7   Beat                  int64  
 8   District              float64
 9   Ward                  float64
 10  Community Area        float64
 11  Year                  int64  
 12  Latitude              float64
 13  Longitude             float64
dtypes: bool(2), float64(5), int64(2), object(5)
memory usage: 213.0+ MB
In [14]:
# A summary statistics of each column
chicago_crime_df.describe(include='all')
Out[14]:
Case Number Date Block Primary Type Location Description Arrest Domestic Beat District Ward Community Area Year Latitude Longitude
count 2278725 2278726 2278726 2278726 2275849 2278726 2278726 2.278726e+06 2.278714e+06 2.094031e+06 2.094459e+06 2.278726e+06 2.254741e+06 2.254741e+06
unique 2278678 1343546 54347 35 198 2 2 NaN NaN NaN NaN NaN NaN NaN
top HK172551 01/01/2008 12:01:00 AM 100XX W OHARE ST THEFT STREET False False NaN NaN NaN NaN NaN NaN NaN
freq 3 81 4878 479375 590957 1675252 1965311 NaN NaN NaN NaN NaN NaN NaN
mean NaN NaN NaN NaN NaN NaN NaN 1.186442e+03 1.129072e+01 2.272764e+01 3.752140e+01 2.009638e+03 4.184209e+01 -8.767161e+01
std NaN NaN NaN NaN NaN NaN NaN 7.026836e+02 6.946692e+00 1.383464e+01 2.153282e+01 6.019724e+00 8.830434e-02 6.073538e-02
min NaN NaN NaN NaN NaN NaN NaN 1.110000e+02 1.000000e+00 1.000000e+00 0.000000e+00 2.001000e+03 3.661945e+01 -9.168657e+01
25% NaN NaN NaN NaN NaN NaN NaN 6.210000e+02 6.000000e+00 1.000000e+01 2.300000e+01 2.004000e+03 4.176866e+01 -8.771379e+01
50% NaN NaN NaN NaN NaN NaN NaN 1.034000e+03 1.000000e+01 2.300000e+01 3.200000e+01 2.009000e+03 4.185578e+01 -8.766597e+01
75% NaN NaN NaN NaN NaN NaN NaN 1.731000e+03 1.700000e+01 3.400000e+01 5.700000e+01 2.014000e+03 4.190668e+01 -8.762823e+01
max NaN NaN NaN NaN NaN NaN NaN 2.535000e+03 3.100000e+01 5.000000e+01 7.700000e+01 2.022000e+03 4.202291e+01 -8.752453e+01
In [15]:
chicago_crime_df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 2278726 entries, 0 to 2278725
Data columns (total 14 columns):
 #   Column                Dtype  
---  ------                -----  
 0   Case Number           object 
 1   Date                  object 
 2   Block                 object 
 3   Primary Type          object 
 4   Location Description  object 
 5   Arrest                bool   
 6   Domestic              bool   
 7   Beat                  int64  
 8   District              float64
 9   Ward                  float64
 10  Community Area        float64
 11  Year                  int64  
 12  Latitude              float64
 13  Longitude             float64
dtypes: bool(2), float64(5), int64(2), object(5)
memory usage: 213.0+ MB

3. Handling missing values

In [16]:
# checking missing value count
chicago_crime_df.isnull().sum()
Out[16]:
Case Number                  1
Date                         0
Block                        0
Primary Type                 0
Location Description      2877
Arrest                       0
Domestic                     0
Beat                         0
District                    12
Ward                    184695
Community Area          184267
Year                         0
Latitude                 23985
Longitude                23985
dtype: int64
In [17]:
# # Missing Values

# Function to calculate missing values by column
def missing_values_table(df):
        # Total missing values
        mis_val = df.isnull().sum()

        # Percentage of missing values
        mis_val_percent = 100 * df.isnull().sum() / len(df)

        # Make a table with the results
        mis_val_table = pd.concat([mis_val, mis_val_percent], axis=1)

        # Rename the columns
        mis_val_table_ren_columns = mis_val_table.rename(
        columns = {0 : 'Missing Values', 1 : '% of Total Values'})

        # Sort the table by percentage of missing descending
        mis_val_table_ren_columns = mis_val_table_ren_columns[
            mis_val_table_ren_columns.iloc[:,1] != 0].sort_values(
        '% of Total Values', ascending=False).round(1)

        # Print some summary information
        print ("Your selected dataframe has " + str(df.shape[1]) + " columns.\n"
            "There are " + str(mis_val_table_ren_columns.shape[0]) +
              " columns that have missing values.")

        # Return the dataframe with missing information
        return mis_val_table_ren_columns
In [18]:
missing_values_table(chicago_crime_df)

Your selected dataframe has 14 columns.
There are 7 columns that have missing values.
Out[18]:
Missing Values % of Total Values
Ward 184695 8.1
Community Area 184267 8.1
Latitude 23985 1.1
Longitude 23985 1.1
Location Description 2877 0.1
District 12 0.0
Case Number 1 0.0
  • Handling Missing Case Number

We have one (1) missing Case Number in the entire dataset. We have decided to drop the row where Case Number is missing

In [19]:
# dropping the missing Case Number column null value
chicago_crime_df.dropna(subset=['Case Number'],inplace=True) # i chose to drop this because the percentage NA is insignificant
In [20]:
chicago_crime_df.isna().sum()
Out[20]:
Case Number                  0
Date                         0
Block                        0
Primary Type                 0
Location Description      2877
Arrest                       0
Domestic                     0
Beat                         0
District                    12
Ward                    184695
Community Area          184267
Year                         0
Latitude                 23984
Longitude                23984
dtype: int64
  • Handling Missing values for the District Column

We have just 12 missing value for the district column which is 0.0% of the entire dataset. We have decided to drop the row where location discription is missing.

In [21]:
# dropping the missing district column null value
chicago_crime_df.dropna(subset=['District'],inplace=True) # i chose to drop this because the percentage NA is insignificant
In [22]:
chicago_crime_df.District.unique()
Out[22]:
array([ 4., 17., 11., 12.,  2., 15., 25., 14.,  8.,  6., 19.,  1.,  7.,
       20., 18.,  3.,  9., 22.,  5., 16., 10., 24., 31.])
In [23]:
chicago_crime_df.isna().sum()
Out[23]:
Case Number                  0
Date                         0
Block                        0
Primary Type                 0
Location Description      2877
Arrest                       0
Domestic                     0
Beat                         0
District                     0
Ward                    184695
Community Area          184267
Year                         0
Latitude                 23984
Longitude                23984
dtype: int64
  • Handling Missing values for the ward Column (Using data imputation)
In [24]:
missing_values_table(chicago_crime_df)

Your selected dataframe has 14 columns.
There are 5 columns that have missing values.
Out[24]:
Missing Values % of Total Values
Ward 184695 8.1
Community Area 184267 8.1
Latitude 23984 1.1
Longitude 23984 1.1
Location Description 2877 0.1
In [25]:
# checking the unique values of the ward column
chicago_crime_df.Ward.unique()
Out[25]:
array([10.,  8., 39., 28., 27., nan, 29., 37., 26., 35., 14.,  1., 17.,
        3., 44., 36.,  4.,  2.,  6.,  7., 40., 42., 20., 13., 11., 34.,
       47., 33.,  9., 21., 45., 25., 24., 15., 16., 19., 43., 30., 23.,
       32., 31., 22., 50., 49.,  5., 46., 12., 41., 38., 18., 48.])
In [26]:
chicago_crime_df.Ward
Out[26]:
0          10.0
1           8.0
2          39.0
3          28.0
4          27.0
           ... 
2278721    42.0
2278722     NaN
2278723     4.0
2278724    42.0
2278725     1.0
Name: Ward, Length: 2278713, dtype: float64
In [27]:
# Using ffill(forward fill) to fill missing value in the ward column
chicago_crime_df.Ward = chicago_crime_df.Ward.ffill() # Last observed value is carried forward
In [28]:
chicago_crime_df.isna().sum()
Out[28]:
Case Number                  0
Date                         0
Block                        0
Primary Type                 0
Location Description      2877
Arrest                       0
Domestic                     0
Beat                         0
District                     0
Ward                         0
Community Area          184267
Year                         0
Latitude                 23984
Longitude                23984
dtype: int64
  • Handling Missing values for the Community Area Column (Using data imputation)
In [29]:
missing_values_table(chicago_crime_df)

Your selected dataframe has 14 columns.
There are 4 columns that have missing values.
Out[29]:
Missing Values % of Total Values
Community Area 184267 8.1
Latitude 23984 1.1
Longitude 23984 1.1
Location Description 2877 0.1
In [30]:
# checking the unique values of the ward column
chicago_crime_df['Community Area'].unique()
Out[30]:
array([46., 47., 12., 26., 28., nan, 25., 23., 22., 27., 63., 24., 71.,
       38.,  6., 19., 55., 36., 32., 69., 43.,  4.,  8., 42., 65., 60.,
       14., 49., 35., 40., 21., 73., 52., 15., 18., 16., 61., 66., 44.,
       45., 67., 74.,  2., 29., 51., 57., 17.,  7., 62., 30., 31., 13.,
       75.,  5.,  1., 39., 53.,  3., 68., 11., 37., 41., 48., 33., 59.,
       58., 76., 54., 10., 20., 34., 77., 64., 56., 70.,  9., 72., 50.,
        0.])
In [31]:
# Using ffill(forward fill) to fill missing value in the Community Area column
chicago_crime_df['Community Area'] = chicago_crime_df['Community Area'].ffill() # Last observed value is carried forward
In [32]:
chicago_crime_df.isna().sum()
Out[32]:
Case Number                 0
Date                        0
Block                       0
Primary Type                0
Location Description     2877
Arrest                      0
Domestic                    0
Beat                        0
District                    0
Ward                        0
Community Area              0
Year                        0
Latitude                23984
Longitude               23984
dtype: int64
  • Handling Missing values for the latitude and Longitude Column (Using Assosciation)
    • The assumption here is that the latitude and longitude readings are points that show exact location, are closer to eachother within a district. if they are, we can group our data by districts and fill up missing Latitide and longitude values by the median or mode.
In [33]:
chicago_crime_df.groupby(['District', 'Longitude', 'Latitude']).size()
Out[33]:
District  Longitude   Latitude 
1.0       -87.776059  41.933944    1
          -87.645624  41.881909    4
          -87.645320  41.885582    2
          -87.645301  41.878079    1
          -87.645121  41.885586    1
                                  ..
31.0      -87.673838  41.776755    1
          -87.631942  41.930460    1
          -87.617227  41.646187    1
          -87.610757  41.884107    1
          -87.562704  41.737240    1
Length: 548193, dtype: int64

There appears to be closeness in longitude and Latitude points when data is grouped based on districts, we can group based on district and fill up by taking the median.

In [34]:
chicago_crime_df['Latitude'] = chicago_crime_df.groupby('District')['Latitude'].transform(lambda x: x.fillna(x.median()))
chicago_crime_df['Longitude'] = chicago_crime_df.groupby('District')['Longitude'].transform(lambda x: x.fillna(x.median()))
In [35]:
chicago_crime_df.isna().sum()
Out[35]:
Case Number                0
Date                       0
Block                      0
Primary Type               0
Location Description    2877
Arrest                     0
Domestic                   0
Beat                       0
District                   0
Ward                       0
Community Area             0
Year                       0
Latitude                   0
Longitude                  0
dtype: int64
  • Handling Missing values for the Location Description

We have just 2877 missing value for the location description which is 0.1% of the entire dataset. We can decide to drop the row where location decription is missing or fill this row with the most crime location.

In [36]:
chicago_crime_df.groupby(['District','Location Description']).size()
Out[36]:
District  Location Description                       
1.0       ABANDONED BUILDING                              18
          AIRPORT BUILDING NON-TERMINAL - SECURE AREA      1
          AIRPORT TRANSPORTATION SYSTEM (ATS)              2
          AIRPORT/AIRCRAFT                                 4
          ALLEY                                          529
                                                        ... 
31.0      RESIDENTIAL YARD (FRONT/BACK)                    1
          SIDEWALK                                         2
          SMALL RETAIL STORE                               2
          STREET                                          12
          VEHICLE NON-COMMERCIAL                           2
Length: 2915, dtype: int64
In [37]:
# dropping the missing location description index row
chicago_crime_df.dropna(subset=['Location Description'],inplace=True)
In [38]:
chicago_crime_df.isna().sum()
Out[38]:
Case Number             0
Date                    0
Block                   0
Primary Type            0
Location Description    0
Arrest                  0
Domestic                0
Beat                    0
District                0
Ward                    0
Community Area          0
Year                    0
Latitude                0
Longitude               0
dtype: int64

3. convert inappropriate data types to the correct one

We should convert the columns to appropriate datatypes for

  1. Alignment with original dataset types.
  2. For performance gains (clearly, using a numpy.float64 when a numpy.int64 is required will lower our memory footprint).
In [39]:
# create a copy of the dataframe to avoid conflict
c_crime_df = chicago_crime_df.copy()
In [40]:
# Including the date format dramatically increases performance.
c_crime_df["Date"] = pd.to_datetime(c_crime_df["Date"], format = "%m/%d/%Y %I:%M:%S %p", utc = True)
c_crime_df["Beat"] = c_crime_df["Beat"].astype(str)
c_crime_df["District"] = c_crime_df["District"].astype(str)
c_crime_df["Ward"] = c_crime_df["Ward"].astype(str)
c_crime_df["Community Area"] = c_crime_df["Community Area"].astype(str)
c_crime_df['Block'] = c_crime_df['Block'].astype(str)

4. Feature Engineering¶

Create new columns for Hour, Date, Block_new, Street, Months, Day, Season and Crime type.

In [41]:
# Create new columns for Months
def extract_and_map_month(date):
    # ensure the input is in datetime format
    # 07/26/2008 02:30:00 PM
    date_column = pd.to_datetime(date, format='%m/%d/%Y %I:%M:%S %p')

    month_names = date_column.dt.month_name()
# dt is an attribute of the date column object
# month_name() is a method
    return month_names
In [42]:
c_crime_df['Month'] = extract_and_map_month(c_crime_df['Date'])

# Group by month and count number of crimes
monthly_crime_counts = c_crime_df.groupby('Month').size()
In [43]:
# Create new columns for Day
def extract_and_map_day(date):
    # ensure the input is in datetime format
    # 07/26/2008 02:30:00 PM
    date_column = pd.to_datetime(date, format='%m/%d/%Y %I:%M:%S %p')

    day_names = date_column.dt.day_name()
# dt is an attribute of the date column object
# day_name() is a method
    return day_names
In [44]:
c_crime_df['Day'] = extract_and_map_day(c_crime_df['Date'])

# Group by day and count number of crimes
daily_crime_counts = c_crime_df.groupby('Day').size()
In [45]:
# Function to define various month and their corresponding season
def get_season(month):
    if month in ['December','January','February']:
        return 'Winter'
    elif month in ['March','April','May']:
        return 'Spring'
    elif month in ['June','July','August']:
        return 'Summer'
    else:
        return 'Fall'
In [46]:
c_crime_df['Season'] = c_crime_df['Month'].apply(get_season)

# Group by season and count number of crimes
seasonal_crime_counts = c_crime_df.groupby('Season').size()
In [47]:
# Creating a new column for hour

# Convert 'Date' column to datetime
c_crime_df['Date'] = pd.to_datetime(c_crime_df['Date'])

# Extract hour from 'Date' and create a new column 'Hour'
c_crime_df['Hour'] = c_crime_df['Date'].dt.hour

Create a Crime Type Table¶

In [48]:
# Classified the crimes based on Theft, Violent Crime, and various other types based off on primary type 
c_crime_df['Crime Type'] = c_crime_df['Primary Type']

c_crime_df['Crime Type'].replace(['THEFT', 'BURGLARY', 'MOTOR VEHICLE THEFT'], 'THEFT', inplace = True)

c_crime_df['Crime Type'].replace(['ASSAULT', 'CRIM SEXUAL ASSAULT', 'SEX OFFENSE', 'CRIMINAL SEXUAL ASSAULT', 'HOMICIDE', 'KIDNAPPING', 'ARSON', 'ROBBERY', 'BATTERY'], 'VIOLENT CRIME', inplace = True)
In [49]:
c_crime_df.head()
Out[49]:
Case Number Date Block Primary Type Location Description Arrest Domestic Beat District Ward Community Area Year Latitude Longitude Month Day Season Hour Crime Type
0 HP485721 2008-07-26 14:30:00+00:00 085XX S MUSKEGON AVE CRIMINAL DAMAGE STREET False False 423 4.0 10.0 46.0 2008 41.739980 -87.555120 July Saturday Summer 14 CRIMINAL DAMAGE
1 JB372830 2018-07-31 10:57:00+00:00 092XX S ELLIS AVE WEAPONS VIOLATION POOL ROOM True False 413 4.0 8.0 47.0 2018 41.726922 -87.599747 July Tuesday Summer 10 WEAPONS VIOLATION
2 HN308568 2007-04-27 10:30:00+00:00 062XX N TRIPP AVE BURGLARY RESIDENCE True False 1711 17.0 39.0 12.0 2007 41.994138 -87.734959 April Friday Spring 10 THEFT
3 JB361368 2018-07-23 08:55:00+00:00 0000X N KEELER AVE ASSAULT NURSING HOME/RETIREMENT HOME False False 1115 11.0 28.0 26.0 2018 41.881217 -87.730590 July Monday Summer 8 VIOLENT CRIME
4 JE297624 2021-07-11 06:40:00+00:00 016XX W HARRISON ST ASSAULT PARKING LOT / GARAGE (NON RESIDENTIAL) False False 1231 12.0 27.0 28.0 2021 41.874174 -87.668082 July Sunday Summer 6 VIOLENT CRIME
In [50]:
c_crime_df.columns
Out[50]:
Index(['Case Number', 'Date', 'Block', 'Primary Type', 'Location Description',
       'Arrest', 'Domestic', 'Beat', 'District', 'Ward', 'Community Area',
       'Year', 'Latitude', 'Longitude', 'Month', 'Day', 'Season', 'Hour',
       'Crime Type'],
      dtype='object')

4. EDA: Data Exploration¶

Non - Graphical¶

  1. What is the total number of crime cases registered by the police department from 2001 to 2022?
In [51]:
print(f"The total number of crime cases registered from 2001 to 2022 is: **{len(c_crime_df)}**")

The total number of crime cases registered from 2001 to 2022 is: **2275836**
  1. What is the number of unique crime cases?
In [52]:
# Number of Unique Case Number.
print(f"Out of {len(c_crime_df)}: {len(c_crime_df['Case Number'].unique())} are unique case numbers")

Out of 2275836: 2275789 are unique case numbers

Observation: Most case numbers are unique case numbers.

  1. What is the number of non-unique cases?
In [53]:
# Out of all cases , how many are non_unique cases ?

non_unique_case_number = c_crime_df[c_crime_df.duplicated('Case Number') | c_crime_df.duplicated('Case Number', keep=False)]['Case Number'].tolist()

print(f"Length of non_unique: {len(non_unique_case_number)} out of {len(c_crime_df)}")

Length of non_unique: 92 out of 2275836
In [54]:
non_unique_case_number_df = c_crime_df[c_crime_df.duplicated('Case Number', keep=False)]
non_unique_case_number_df.head() # Saving non_unique case number dataframe for future use.
Out[54]:
Case Number Date Block Primary Type Location Description Arrest Domestic Beat District Ward Community Area Year Latitude Longitude Month Day Season Hour Crime Type
25445 HM763064 2006-12-08 15:14:00+00:00 005XX W MADISON ST HOMICIDE OFFICE True False 111 1.0 42.0 28.0 2006 41.881884 -87.640060 December Friday Winter 15 VIOLENT CRIME
58104 JB483810 2018-10-21 02:39:00+00:00 001XX N LOCKWOOD AVE HOMICIDE AUTO False False 1523 15.0 28.0 25.0 2018 41.883604 -87.757629 October Sunday Fall 2 VIOLENT CRIME
74485 HK172551 2004-02-08 23:44:00+00:00 034XX W MADISON ST HOMICIDE CHURCH False False 1123 11.0 28.0 27.0 2004 41.880909 -87.711843 February Sunday Winter 23 VIOLENT CRIME
80580 HV620483 2012-12-31 16:49:00+00:00 042XX W 21ST PL HOMICIDE APARTMENT True True 1012 10.0 24.0 29.0 2012 41.852505 -87.730687 December Monday Winter 16 VIOLENT CRIME
126290 HR146859 2009-01-31 01:10:00+00:00 039XX W ARGYLE ST HOMICIDE APARTMENT True False 1712 17.0 39.0 14.0 2009 41.971903 -87.726290 January Saturday Winter 1 VIOLENT CRIME

Observation: There are 2275789 unique values of case number out of 2275836 , it means there are 2275789 cases registered with a single crime the rest 92 are the case numbers which have commited multiple crimes.

  1. Out of total case registered how many are arrested and how many are not?
In [55]:
# Out of total case registered how many are arrested and how many are not?

arrested = c_crime_df['Arrest'].where(c_crime_df['Arrest'] == True).count()
not_arrested = c_crime_df['Arrest'].where(c_crime_df['Arrest'] != True).count()

print(f"Out of {len(c_crime_df)},\n{arrested} are Arrested and \n{not_arrested} are not Arrested.")

Out of 2275836,
603465 are Arrested and 
1672371 are not Arrested.

Observation: Out of total crime cases received, 27% of the Criminals are arrested, 73% were not.

  1. How many unique Primary type is available and what is the count for each crime from 2001-2022?
In [56]:
# Group by Primary type and calculate count
crime_count = c_crime_df['Primary Type'].unique()
print(f'The Number of unique Primary crimes are {len(crime_count)} and the count for each crime are:')
count_for_each_crime = pd.DataFrame(c_crime_df.groupby('Primary Type').size().sort_values(ascending=False).rename('Count').reset_index())
count_for_each_crime

The Number of unique Primary crimes are 35 and the count for each crime are:
Out[56]:
Primary Type Count
0 THEFT 479153
1 BATTERY 418647
2 CRIMINAL DAMAGE 259959
3 NARCOTICS 222861
4 ASSAULT 147662
5 OTHER OFFENSE 141627
6 BURGLARY 125161
7 MOTOR VEHICLE THEFT 105852
8 DECEPTIVE PRACTICE 96707
9 ROBBERY 85694
10 CRIMINAL TRESPASS 63498
11 WEAPONS VIOLATION 30210
12 PROSTITUTION 20795
13 OFFENSE INVOLVING CHILDREN 16320
14 PUBLIC PEACE VIOLATION 15510
15 SEX OFFENSE 8806
16 CRIM SEXUAL ASSAULT 8363
17 INTERFERENCE WITH PUBLIC OFFICER 5494
18 LIQUOR LAW VIOLATION 4470
19 GAMBLING 4345
20 ARSON 3886
21 HOMICIDE 3597
22 KIDNAPPING 2172
23 CRIMINAL SEXUAL ASSAULT 1589
24 INTIMIDATION 1397
25 STALKING 1338
26 CONCEALED CARRY LICENSE VIOLATION 305
27 OBSCENITY 224
28 PUBLIC INDECENCY 58
29 NON-CRIMINAL 48
30 HUMAN TRAFFICKING 32
31 OTHER NARCOTIC VIOLATION 31
32 NON - CRIMINAL 12
33 RITUALISM 11
34 NON-CRIMINAL (SUBJECT SPECIFIED) 2

Observation:

The number of unique primary types available in the dataset from 2001 to 2022 are 35, The count for each crime type during this period are shown above with Theft Being the highest with a count of 479153. This analysis provides insights into the prevalence of different types of crimes over the specified time frame, particularly focusing on the occurrence of theft, which is the highest among all primary types.

  1. what is the number of crime per year?
In [57]:
# Count the number of crimes per year
crime_counts_per_year = c_crime_df['Year'].value_counts().sort_index().rename('Count').reset_index()
print(f'The amount of crimes per year are as follow: ')
crime_counts_per_year

The amount of crimes per year are as follow:
Out[57]:
Year Count
0 2001 146292
1 2002 145441
2 2003 142708
3 2004 140802
4 2005 135737
5 2006 134769
6 2007 130754
7 2008 127883
8 2009 118376
9 2010 111077
10 2011 105639
11 2012 100724
12 2013 91802
13 2014 82597
14 2015 79183
15 2016 80872
16 2017 80312
17 2018 80702
18 2019 77719
19 2020 63193
20 2021 62284
21 2022 36970

Observation:

This analysis shows that highest crime rate happened in 2001 and could be the reason behind the birth of the crime data collection methods and reporting practices, which may be part of a cyclical pattern or a one-time anomaly. In order to provide a conclusion, we need to examine data from previous years to provide insight into the long-term trend and whether the increase in 2001 was part of a larger cycle.

  1. what is the number of crime per month?
In [58]:
# Reorder the days of the week
month_order = ['January', 'February', 'March', 'April', 'May', 'June', 'July','August','September','October','November','December']

# Count the number of crimes per Month
crime_counts_per_month = c_crime_df['Month'].value_counts().rename('Count').reindex(month_order)
print(f'The amount of crimes per month are as follow: ')
crime_counts_per_month

The amount of crimes per month are as follow:
Out[58]:
Month
January      179894
February     158349
March        188532
April        187239
May          204425
June         204809
July         213452
August       206089
September    193278
October      195487
November     176376
December     167906
Name: Count, dtype: int64

Observation:

The observation regarding the number of crimes per month where July saw the highest crime count suggests a potential trend in criminal activity throughout the years. This can be as a cause of Seasonal Variations because July falls within the summer season in many regions, and warmer weather conditions may contribute to increased outdoor activities, larger gatherings, and higher rates of certain types of crimes such as theft, burglary, and vandalism. Also, summer months like July often coincide with school vacations, holidays, and festivals, leading to changes in routines and behaviors among individuals, potentially affecting crime dynamics.

  1. what is the number of crime by season?
In [59]:
# Count the number of crimes per Seasons
crime_counts_per_season = c_crime_df['Season'].value_counts().sort_index().rename('Count').reset_index()
print(f'The Seasonal Crime rate are as follow: ')
crime_counts_per_season

The Seasonal Crime rate are as follow:
Out[59]:
Season Count
0 Fall 565141
1 Spring 580196
2 Summer 624350
3 Winter 506149

Observation:

Based on the analysis, Summer is identified as the season with the highest crime count, it suggests that crime rates tend to peak during the summer months (June, July, August). This observation aligns with the common trend observed in many regions where warmer weather and longer days may contribute to increased outdoor activities, which could potentially lead to higher crime rates.

  1. what is the number of crime per day?
In [60]:
# Reorder the days of the week
day_order = ['Monday', 'Tuesday', 'Wednesday', 'Thursday', 'Friday', 'Saturday', 'Sunday']

# Count the number of crimes per day
crime_counts_per_day = c_crime_df['Day'].value_counts().rename('Count').reindex(day_order)
print(f'The daily Crime rate are as follow: ')
crime_counts_per_day

The daily Crime rate are as follow:
Out[60]:
Day
Monday       321298
Tuesday      325467
Wednesday    327674
Thursday     323885
Friday       342109
Saturday     326139
Sunday       309264
Name: Count, dtype: int64

Observation

The observation is that Friday had the highest number of crime occurrences compared to other days of the week. This indicates a potential pattern or trend in criminal activity, with Fridays being more prone to criminal incidents in the Chicago area. Such a trend may be:

  1. Increased social activities: Fridays are often associated with the end of the workweek and the beginning of the weekend, leading to increased social gatherings, nightlife, and events. This higher level of activity may create opportunities for certain types of crimes, such as thefts, assaults, or disturbances.
  2. Alcohol consumption: Friday nights often involve alcohol consumption and partying, which can lead to heightened emotions and risky behavior, potentially resulting in incidents like fights, public intoxication, or DUI offenses.
  3. Routine disruptions: The transition from the workweek to the weekend may disrupt regular routines and patterns, making it easier for criminal elements to take advantage of the situation.
  1. What are the top 10 most common Primary crime types?
In [61]:
# Top 10 primary type of crimes
common_crime = c_crime_df['Primary Type'].value_counts().head(10)
print('The top 10 most common primary types are ')
common_crime

The top 10 most common primary types are
Out[61]:
Primary Type
THEFT                  479153
BATTERY                418647
CRIMINAL DAMAGE        259959
NARCOTICS              222861
ASSAULT                147662
OTHER OFFENSE          141627
BURGLARY               125161
MOTOR VEHICLE THEFT    105852
DECEPTIVE PRACTICE      96707
ROBBERY                 85694
Name: count, dtype: int64

Observation:

Among the top 10 most common primary crime types, theft stands out as the crime type with the highest crime count. This observation suggests that theft is a prevalent issue in the Chicago area, based on the available crime data. The high frequency of theft incidents underscores the importance of addressing theft-related crimes through targeted prevention strategies, increased law enforcement efforts, community engagement, and public awareness campaigns.

  1. what are the top 10 most common location description for crimes?
In [62]:
# Top 10 crime Location
common_location = c_crime_df['Location Description'].value_counts().head(10)
print('The top 10 most common crime Location are: ')
common_location

The top 10 most common crime Location are:
Out[62]:
Location Description
STREET                            590954
RESIDENCE                         384471
APARTMENT                         253114
SIDEWALK                          216275
OTHER                              81243
PARKING LOT/GARAGE(NON.RESID.)     61109
ALLEY                              50652
SCHOOL, PUBLIC, BUILDING           43678
SMALL RETAIL STORE                 42803
RESIDENCE-GARAGE                   40478
Name: count, dtype: int64

Observation:

Among the top 10 most common location descriptions for crimes, "STREET" appears to be the most frequent location where crimes occur. This observation suggests that street-related incidents are prevalent in the dataset and highlights the significance of outdoor public spaces as settings for criminal activity.

  1. what are the top 10 most common crimes in 2021?
In [63]:
# Top 10 crimes in 2021
crime_2021 = c_crime_df[c_crime_df['Year'] == 2021]
print("Top 10 most crime in 2021: ")
crime_2021['Primary Type'].value_counts().head(10)

Top 10 most crime in 2021:
Out[63]:
Primary Type
THEFT                  12313
BATTERY                12017
CRIMINAL DAMAGE         7571
ASSAULT                 6134
DECEPTIVE PRACTICE      4905
OTHER OFFENSE           4120
MOTOR VEHICLE THEFT     3163
WEAPONS VIOLATION       2752
ROBBERY                 2407
BURGLARY                1978
Name: count, dtype: int64

Observation:

In 2021, among the top 10 most common crimes, "Theft" had the highest count. This observation suggests that theft was the most prevalent type of crime in Chicago during that year, based on the data available.

  1. How many crimes were reported each month in 2021 ?
In [64]:
# How many crimes were reported each month in 2021 ?
monthly_crime_2021 = crime_2021['Month'].value_counts().reindex(month_order)
print('Number of crimes reported each month in 2021:')
monthly_crime_2021

Number of crimes reported each month in 2021:
Out[64]:
Month
January      4848
February     3851
March        4839
April        4609
May          5478
June         5585
July         5630
August       5590
September    5721
October      5809
November     5127
December     5197
Name: count, dtype: int64

Observation:

The observation indicates that there was a notable increase in reported crimes in October 2021 compared to other months in the dataset. An increase in reported crimes in October could be attributed to specific events or occurrences during that month, such as festivals, holidays, or social gatherings that may lead to heightened criminal activity.

  1. what is the top 10 days with the most crime count ?
In [65]:
# Group by date and count the number of cases
cases_by_date = c_crime_df.groupby('Date').size().reset_index(name='no_of_cases')

# Extract day, month, and year from the 'Date' column
cases_by_date['Day'] = cases_by_date['Date'].dt.day
cases_by_date['Month'] = cases_by_date['Date'].dt.month
cases_by_date['Year'] = cases_by_date['Date'].dt.year

# Select the top 10 days with the highest number of cases
top_10_days = cases_by_date.nlargest(10, 'no_of_cases')

# Select columns for the table
top_10_days_table = top_10_days[['Day', 'Month', 'Year', 'no_of_cases']].set_index('no_of_cases')
top_10_days_table
Out[65]:
Day Month Year
no_of_cases
81 1 1 2008
79 1 1 2007
68 1 1 2007
65 1 1 2003
64 1 1 2004
63 1 1 2005
62 1 1 2006
59 1 1 2001
59 1 1 2003
59 1 1 2004

Observation:

The observation indicates that the highest crime count often occurs on the first day of the first month of a new year. This pattern could be attributed to various factors such as:

  1. New Year's Celebrations: The first day of the year, especially during New Year's Eve celebrations, may witness increased incidents of certain types of crimes such as public disturbances, disorderly conduct, and alcohol-related offenses.

  2. Large Gatherings: New Year's Eve events, parties, and festivities often draw large crowds, increasing the likelihood of certain types of crimes such as thefts, assaults, and property damage.

  1. Crimes with the most Arrest
In [66]:
# Crimes with the most arrests
c_crime_df[c_crime_df['Arrest'] == True].groupby('Crime Type').size().sort_values(ascending = False)
Out[66]:
Crime Type
NARCOTICS                            221528
VIOLENT CRIME                        138992
THEFT                                 70394
CRIMINAL TRESPASS                     44742
OTHER OFFENSE                         25026
WEAPONS VIOLATION                     22689
PROSTITUTION                          20713
CRIMINAL DAMAGE                       17629
DECEPTIVE PRACTICE                    13747
PUBLIC PEACE VIOLATION                 9926
INTERFERENCE WITH PUBLIC OFFICER       5052
LIQUOR LAW VIOLATION                   4431
GAMBLING                               4308
OFFENSE INVOLVING CHILDREN             3314
CONCEALED CARRY LICENSE VIOLATION       298
INTIMIDATION                            214
STALKING                                204
OBSCENITY                               166
PUBLIC INDECENCY                         58
OTHER NARCOTIC VIOLATION                 22
NON-CRIMINAL                              6
HUMAN TRAFFICKING                         3
NON - CRIMINAL                            2
RITUALISM                                 1
dtype: int64

Observation:

The observation indicates that narcotics-related crimes have the highest arrest rate, which can be due to the following:

  1. Law Enforcement Focus: The high number of arrests in narcotics-related crimes suggests a significant focus by law enforcement agencies on combating drug-related activities. This focus may involve specialized units, task forces, and strategies aimed at apprehending individuals involved in drug trafficking, distribution, or possession.

  2. Criminal Justice Prioritization: The prioritization of narcotics-related crimes for arrest reflects the recognition of the significant impact of drug-related activities on public safety, health, and community well-being. Addressing narcotics-related offenses may be seen as a priority within the criminal justice system due to their potential to fuel other types of criminal activity and societal harm.

  1. Count of Arrest and Non arrest for each Crime type.
In [67]:
# Count of Arrest and Non arrest for each Crime type.
c_crime_df.groupby(['Crime Type','Arrest']).agg({'Arrest': 'count'})
Out[67]:
Arrest
Crime Type Arrest
CONCEALED CARRY LICENSE VIOLATION False 7
True 298
CRIMINAL DAMAGE False 242330
True 17629
CRIMINAL TRESPASS False 18756
True 44742
DECEPTIVE PRACTICE False 82960
True 13747
GAMBLING False 37
True 4308
HUMAN TRAFFICKING False 29
True 3
INTERFERENCE WITH PUBLIC OFFICER False 442
True 5052
INTIMIDATION False 1183
True 214
LIQUOR LAW VIOLATION False 39
True 4431
NARCOTICS False 1333
True 221528
NON - CRIMINAL False 10
True 2
NON-CRIMINAL False 42
True 6
NON-CRIMINAL (SUBJECT SPECIFIED) False 2
OBSCENITY False 58
True 166
OFFENSE INVOLVING CHILDREN False 13006
True 3314
OTHER NARCOTIC VIOLATION False 9
True 22
OTHER OFFENSE False 116601
True 25026
PROSTITUTION False 82
True 20713
PUBLIC INDECENCY True 58
PUBLIC PEACE VIOLATION False 5584
True 9926
RITUALISM False 10
True 1
STALKING False 1134
True 204
THEFT False 639772
True 70394
VIOLENT CRIME False 541424
True 138992
WEAPONS VIOLATION False 7521
True 22689
  1. Percentage of Arrest and Non arrest for each Crime type.
In [68]:
# Group by crime type and arrest, and count the occurrences
crime_type_arrest_counts = c_crime_df.groupby(['Crime Type', 'Arrest']).size()

# Calculate total counts for each crime type
total_counts = crime_type_arrest_counts.groupby('Crime Type').sum()

# Calculate the percentage of arrests and non-arrests for each crime type
arrest_percentage = ((crime_type_arrest_counts.loc[:, True] / total_counts) * 100).round(1)
non_arrest_percentage =((crime_type_arrest_counts.loc[:, False] / total_counts) * 100).round(1)

# Combine the percentages into a DataFrame
percentage_df = pd.DataFrame({'Arrest Percentage': arrest_percentage, 'Non-Arrest Percentage': non_arrest_percentage})

# Display the DataFrame
percentage_df
Out[68]:
Arrest Percentage Non-Arrest Percentage
Crime Type
CONCEALED CARRY LICENSE VIOLATION 97.7 2.3
CRIMINAL DAMAGE 6.8 93.2
CRIMINAL TRESPASS 70.5 29.5
DECEPTIVE PRACTICE 14.2 85.8
GAMBLING 99.1 0.9
HUMAN TRAFFICKING 9.4 90.6
INTERFERENCE WITH PUBLIC OFFICER 92.0 8.0
INTIMIDATION 15.3 84.7
LIQUOR LAW VIOLATION 99.1 0.9
NARCOTICS 99.4 0.6
NON - CRIMINAL 16.7 83.3
NON-CRIMINAL 12.5 87.5
NON-CRIMINAL (SUBJECT SPECIFIED) NaN 100.0
OBSCENITY 74.1 25.9
OFFENSE INVOLVING CHILDREN 20.3 79.7
OTHER NARCOTIC VIOLATION 71.0 29.0
OTHER OFFENSE 17.7 82.3
PROSTITUTION 99.6 0.4
PUBLIC INDECENCY 100.0 NaN
PUBLIC PEACE VIOLATION 64.0 36.0
RITUALISM 9.1 90.9
STALKING 15.2 84.8
THEFT 9.9 90.1
VIOLENT CRIME 20.4 79.6
WEAPONS VIOLATION 75.1 24.9

Observation

The observation regarding the Percentage of Arrest and Non-arrest for each Crime type, particularly noting that "Public Indecency" has a 100% arrest rate, suggests some insights into law enforcement practices and the nature of the crime itself. It shows that:

  1. Effective Enforcement: The 100% arrest rate for "Public Indecency" indicates that law enforcement agencies are highly effective in apprehending individuals involved in this type of crime. It may suggest that public indecency is a highly visible or easily detectable offense, allowing law enforcement to respond promptly and make arrests in nearly all cases.

  2. Zero Tolerance Policy: A 100% arrest rate for public indecency could reflect a zero-tolerance policy adopted by law enforcement agencies towards this type of behavior. Such policies often prioritize immediate intervention and enforcement actions, resulting in a high arrest rate for offenders.

  3. Severity of the Offense: Public indecency may be considered a serious offense with clear legal boundaries, leading to swift law enforcement action and arrest of perpetrators. The perceived impact on public safety or community well-being may drive authorities to prioritize enforcement efforts and ensure that offenders are held accountable.

  4. Cultural or Societal Norms: The 100% arrest rate for public indecency may also reflect societal norms or community expectations regarding public behavior. In cultures where public decency and modesty are highly valued, authorities may respond assertively to instances of indecent exposure or behavior in public spaces.

Graphical EDA¶

In [69]:
def plot_counts(serie, title):
    df = pd.DataFrame(serie.value_counts())
    df.columns = ["Freq"]
    df["Type"] = df.index
    fig = px.line(df, x="Type", y="Freq", markers=True, line_shape='linear')
    fig.update_traces(texttemplate="%{y:.2s}", textposition="top center")
    fig.update_layout(uniformtext_minsize=8, uniformtext_mode="hide")
    fig.update_layout(title_text=title)
    fig.show()
In [70]:
# Trend of Crimes in Chicago from 2001 to 2022
plot_counts(c_crime_df["Year"], "Trend of Crimes in Chicago from 2001 to 2022")

Observations:

  1. The crime rate is dropping overall.
  2. In 2001,number of crime incidents were close to 150k.
  3. In 2022 ,it's estimate to be 36k , so it's a major drop in crime.
  4. By 2023 we expect a major drop down to an estimate of 10k
In [71]:
# Monthly crime trend

# Group by month and count occurrences
monthly_crime_counts = c_crime_df['Date'].dt.month.value_counts().sort_index()

# Create DataFrame
df = pd.DataFrame({'Month': monthly_crime_counts.index, 'Frequency': monthly_crime_counts.values})

# Rename month numbers to month names
month_names = ['January', 'February', 'March', 'April', 'May', 'June', 'July', 'August', 'September', 'October', 'November', 'December']
df['Month'] = df['Month'].map(lambda x: month_names[x-1])

# Plot
fig = px.line(df, x='Month', y='Frequency', markers=True, line_shape='linear')
fig.update_traces(texttemplate="%{y:.2s}", textposition="top center")
fig.update_layout(title='Monthly Distribution of Crimes in Chicago (January to December)',
                  xaxis_title='Month',
                  yaxis_title='Number of Crimes',
                  uniformtext_minsize=8,
                  uniformtext_mode="hide")
fig.show()

Observation:

The monthly crime trend in the dataset shows a distinct pattern, with July standing out as the month with the highest peak of crime occurrences, while February had the least. This observation suggests a potential seasonal effect, with higher crime rates during the summer months, possibly due to increased outdoor activities and gatherings. In contrast, the lower crime rates in February might be attributed to various factors such as colder weather conditions, reduced outdoor activities, or fewer opportunities for certain types of crimes. Understanding these seasonal variations in crime trends is crucial for law enforcement agencies and policymakers to allocate resources effectively and implement targeted crime prevention strategies throughout the year.

In [72]:
# Seasonal Trend of Crimes in Chicago
plot_counts(c_crime_df["Season"], "Seasonal Crime Trend")

Observation:

The observation of a seasonal trend of crimes in Chicago, with summer having the highest peak of crime and winter having the least, aligns with common patterns observed in many urban areas. The increase in crime during summer months can be attributed to various factors, including warmer weather and longer daylight hours. These conditions may lead to increased outdoor activities, larger crowds gathering in public spaces, and heightened social interactions, providing more opportunities for criminal activity.

In [73]:
# Daily crime trend 

# Group by day of the week and count the number of crimes
daily_crime_counts = c_crime_df['Day'].value_counts().sort_index()

# Create a DataFrame from the counts
df = pd.DataFrame({'Day': daily_crime_counts.index, 'Number of Crimes': daily_crime_counts.values})

# Define the order of days (Monday to Sunday)
days_order = ['Monday', 'Tuesday', 'Wednesday', 'Thursday', 'Friday', 'Saturday', 'Sunday']

# Create a line plot
fig = px.line(df, x='Day', y='Number of Crimes', category_orders={'Day': days_order}, markers=True, line_shape='linear')

# Customize the plot
fig.update_traces(texttemplate="%{y:.2s}", textposition="top center")
fig.update_layout(title="Daily Distribution of Crimes", xaxis_title="Day of the Week", yaxis_title="Number of Crimes")
fig.show()

Observation:

The daily crime trend in which Friday had the highest peak of crime and Sunday had the least suggests a pattern in criminal activity throughout the week. Here are some observations based on this trend:

  1. Weekend Peak: The noticeable peak in crime on Fridays, followed by a gradual decrease over the weekend, indicates a potential association between criminal activity and weekend social dynamics. Fridays often mark the beginning of the weekend when people may engage in leisure activities, social gatherings, or nightlife, providing opportunities for certain types of crimes such as theft, assault, or public disturbances.

  2. Sunday Lull: The lower crime rate on Sundays suggests a contrast to Fridays, possibly due to decreased social activities, business closures, and reduced foot traffic in public areas. Individuals may be more likely to stay at home or engage in quieter, family-oriented activities on Sundays, leading to fewer opportunities for criminal incidents.

  3. Routine Activities Theory: The daily crime trend aligns with the Routine Activities Theory, which suggests that crime occurs when three elements converge: motivated offenders, suitable targets, and the absence of capable guardians. On Fridays, with increased social activities and larger crowds, there may be more opportunities for crimes to occur. In contrast, Sundays, with reduced social interactions and fewer people out in public spaces, may serve as a deterrent to criminal activity.

In [74]:
# Plot the top 10 days with the most crime counts each day using Plotly
fig = px.bar(top_10_days, x='Date', y='no_of_cases', labels={'Date': 'Date', 'no_of_cases': 'Number of Cases'},
             title='Top 10 Days with the Most Crime Counts Each Day')
fig.update_xaxes(type='category')
fig.show()

Observation: The average Number of cases per day could be less than 60

Top 6 Most reported Crime¶

In [75]:
crime_count = c_crime_df['Crime Type'].value_counts()[:6]
explode = (0.1,0,0,0,0,0)
plt.style.use('ggplot')
plt.title('Percentage of Most Reported Crime')
crime_count.plot.pie(explode = explode, autopct = '%.2f%%', shadow=False, startangle = 90, figsize = (8,8))
plt.axis('equal')
Out[75]:
(-1.1914171612804971,
 1.1043527692389095,
 -1.1024596006774912,
 1.1516517019010946)
No description has been provided for this image

Observation

  1. 33.63% of the overall crime in chicago is attributed to theft
  2. This figure is strongly followed by Violent crimes at 32.22%
  3. Violent crime here is inclusive of Homicide, sexual violence, Battery, Assault, Robbery and Kidnapping
  4. Theft here is inclusive of both property theft such as vehicle theft and petty theft.

Top 6 Arrest per Crime Type

In [76]:
#Top 6 Arrest per crime type

arrest_count = c_crime_df[c_crime_df['Arrest'] == True].groupby('Crime Type').size().sort_values(ascending = False)[:6]
explode = (0.1,0,0,0,0,0)
plt.style.use('ggplot')
plt.title('Top 6 Arrests per crime')
arrest_count.plot.pie(explode = explode, autopct = '%.2f%%', shadow=True, startangle = 90, figsize = (7,7))
plt.axis('equal')
Out[76]:
(-1.2019553794941775,
 1.1048549920683892,
 -1.1011935131129749,
 1.1250658327623646)
No description has been provided for this image

Observation

  1. Narcotics crime has the highest percentage of arrests at 42.33% compared to theft and violent crimes
  2. This is followed by violent crime and theft at 26.56% and 13.45% respectively
  3. This shows that there might be many unsolved violent crime and theft cases in Chicago

Top 10 common crime by type

In [77]:
# We try to indentify the trend and highpoints or lowpoints.
# Function to plot the serie value_counts
def plot_counts2(serie, title):
    df = pd.DataFrame(serie.value_counts()[:10])
    df.columns = ["Freq"]
    df["Type"] = df.index
    fig = px.pie(df, values='Freq', names='Type', title=title)
    fig.show()


plot_counts2(c_crime_df["Crime Type"], "Top 10 common crime by type")

Observation:

THEFT, Violent and CRIMINAL DAMAGE represents more the 62% of the crimes.

Top 10 Common location for crimes¶

In [78]:
def plot_counts3(serie, title):
    # Create a DataFrame with the top 10 values
    df = pd.DataFrame(serie.value_counts()[:10])
    df.columns = ["Freq"]
    df["Type"] = df.index
    
    # Plot the bar chart
    fig = px.bar(df, x='Type', y='Freq', title=title)
    fig.show()
In [79]:
# Where does crimes happens the most?
plot_counts3(c_crime_df["Location Description"], "Top 10 Location of Crimes")
In [80]:
# Top 10 ward with the most crime in chicago
plot_counts3(c_crime_df["Community Area"], "Top 10 community with Most Crime in Chicago")
In [ ]:
In [81]:
# Top 10 ward with the most crime in chicago
plot_counts3(c_crime_df["Ward"], "Top 10 Ward with Most Crime in Chicago")
In [82]:
# Top 10 ward with the most crime in chicago
plot_counts3(c_crime_df["Block"], "Top 10 Block with Most Crime in Chicago")

Observation:

After analyzing the location descriptions, it was found that STREETS are one of the most common locations where crimes occur in the dataset. Further analysis revealed that "100XX IN OHARE Street" had the highest peak of crime incidents, indicating that it is a hotspot for criminal activity. This observation suggests that targeted interventions and enhanced policing efforts may be needed to address crime issues in that area.

Top - 10 communities with the most crime in 2021¶

In [83]:
#  Filter the dataset for the year 2021
chicago_crime_2021 = c_crime_df[c_crime_df['Year'] == 2021]

# Group by community and count the number of crimes
crime_by_community = chicago_crime_2021['Community Area'].value_counts().sort_values(ascending=False)

# Select the top 10 communities with the most crime
top_10_communities = crime_by_community.head(10)

# Create a DataFrame from top 10 communities data
top_10_df = pd.DataFrame({'Community': top_10_communities.index, 'Number of Crimes': top_10_communities.values})

# Plot the top 10 communities with the most crime using a bar plot
fig = px.bar(top_10_df, x='Number of Crimes', y='Community', orientation='h', text='Number of Crimes')
fig.update_traces(texttemplate='%{text}', textposition='inside')
fig.update_layout(title='Top 10 Communities with the Most Crime in 2021', xaxis_title='Number of Crimes', yaxis_title='Community')
fig.show()

Top 5 Primary Crime Types and Their Locations in Chicago (2021)¶

In [84]:
# Filter the dataset to include only incidents that occurred in 2021
crime_2021 = c_crime_df[c_crime_df['Year'] == 2021]

# Get the top 10 crime types in 2021
top_crime_types_2021 = crime_2021['Primary Type'].value_counts().head().index

# Initialize subplots with multiple rows and columns
fig = make_subplots(rows=len(top_crime_types_2021), cols=1, subplot_titles=top_crime_types_2021)

# Iterate over each top crime type
for i, crime_type in enumerate(top_crime_types_2021, start=1):
    # Filter the dataset for the current crime type
    crime_type_df = crime_2021[crime_2021['Primary Type'] == crime_type]
    
    # Count the occurrences of each location for the current crime type
    location_counts = crime_type_df['Location Description'].value_counts().head()
    
    # Create a bar plot for the current crime type and its top 10 locations
    fig.add_trace(go.Bar(x=location_counts.values, y=location_counts.index, orientation='h',
                         name=crime_type), row=i, col=1)

# Update layout
fig.update_layout(height=1000, title="Top Crime Types and Their Locations in Chicago (2021)",
                  xaxis=dict(title="Number of Incidents"), yaxis=dict(title="Location Description"))

# Show plot
fig.show()

Top - 10 Location where theft normally occurs¶

In [85]:
# Filter the dataset to include only theft-related incidents
theft_df = c_crime_df[c_crime_df['Primary Type'] == 'THEFT']

# Count the occurrences of theft incidents at each location
theft_locations_count = theft_df['Location Description'].value_counts()

# Get the top 10 locations
top_10_theft_locations = theft_locations_count.head(10)

# Create a DataFrame for plotting
df = pd.DataFrame({'Location Description': top_10_theft_locations.index, 'Number of Thefts': top_10_theft_locations.values})

# Plot the top 10 theft locations using a bar plot
fig = px.bar(df, x='Number of Thefts', y='Location Description', orientation='h', text='Number of Thefts')
fig.update_traces(textposition='outside')
fig.update_layout(title='Top 10 Locations for Theft in Chicago', xaxis_title='Number of Thefts', yaxis_title='Location Description')
fig.show()

Top - 10 Location for theft in Chicago in 2021¶

In [86]:
c_crime_df.columns
Out[86]:
Index(['Case Number', 'Date', 'Block', 'Primary Type', 'Location Description',
       'Arrest', 'Domestic', 'Beat', 'District', 'Ward', 'Community Area',
       'Year', 'Latitude', 'Longitude', 'Month', 'Day', 'Season', 'Hour',
       'Crime Type'],
      dtype='object')
In [87]:
# Filter the dataset to include only theft-related incidents in 2021
theft_2021_df = c_crime_df[(c_crime_df['Primary Type'] == 'THEFT') & (c_crime_df.Year == 2021)]

# Count the occurrences of theft incidents at each location
theft_locations_count = theft_2021_df['Location Description'].value_counts()

# Get the top 10 locations
top_10_theft_locations = theft_locations_count.head(10)

# Create a DataFrame for plotting
df = pd.DataFrame({'Location Description': top_10_theft_locations.index, 'Number of Thefts': top_10_theft_locations.values})

# Plot the top 10 theft locations using a bar plot
fig = px.bar(df, x='Number of Thefts', y='Location Description', orientation='h', text='Number of Thefts')
fig.update_traces(textposition='outside')
fig.update_layout(title='Top 10 Locations for Theft in Chicago in 2021', xaxis_title='Number of Thefts', yaxis_title='Location Description')
fig.show()

Number of crimes per Year x Months

Let's see the number of crimes per Year and Months...

In [88]:
# first we convert the date column to datetime data type.

c_crime_df['Date'] = pd.to_datetime(chicago_crime_df['Date'],format = "%m/%d/%Y %I:%M:%S %p")
c_crime_df.set_index("Date", inplace=True)
In [89]:
cases = pd.DataFrame(c_crime_df.resample("M").size(), columns=["Number of cases"])
cases["Month"] = cases.index.month.astype(str)
cases["Year"] = cases.index.year.astype(str)
cases["Year-Month"] = cases["Year"].str.cat(cases["Month"].str.zfill(2), sep="-")
cases.head()
Out[89]:
Number of cases Month Year Year-Month
Date
2001-01-31 11479 1 2001 2001-01
2001-02-28 10134 2 2001 2001-02
2001-03-31 12129 3 2001 2001-03
2001-04-30 11965 4 2001 2001-04
2001-05-31 12566 5 2001 2001-05

BY HEATMAP

In [90]:
# Pivot the DataFrame to get crime counts for each month and year
crime_table = c_crime_df.pivot_table(index=c_crime_df.index.month_name(), columns=c_crime_df.index.year, aggfunc='size', fill_value=0)

# Reorder the index to match month order
month_order = ['January', 'February', 'March', 'April', 'May', 'June', 'July', 'August', 'September', 'October', 'November', 'December']
crime_table = crime_table.reindex(month_order)

# Plot heatmap
plt.figure(figsize=(20, 10))
sns.heatmap(crime_table, annot=True, fmt="d", linewidths=0.5, cmap="Blues")
plt.title('Number of Crimes in Chicago per Year x Month')
plt.xlabel('Year')
plt.ylabel('Month')
plt.show()
No description has been provided for this image

The years between 2001 and 2012 were a tough time, the year 2008 was the worst with a mean of 71004 crimes per month, and the best year was the 2015 with a mean of 21916 crimes per month.

BY BAR

In [91]:
fig = px.bar(
    cases,
    x="Year-Month",
    y="Number of cases",
    hover_data=["Year", "Month", "Number of cases"],
    color="Number of cases",
    height=600,
    width=1400,
    color_continuous_scale=px.colors.sequential.Blugrn

)
fig.update_layout(title_text='Crime counts per year-month in bars')
fig.show()

Similar graph but using bars to visualize though noisy but can be used. Again the 'tough time' during 2001 until 2012.

Are there specific location that crimes used to happen?¶

Using Map plot

In [92]:
map_marks = chicago_crime_df[['Latitude', 'Longitude']].copy()
map_marks['Year'] =  pd.to_datetime(chicago_crime_df.Date, format="%m/%d/%Y %I:%M:%S %p").dt.year
map_marks.head()
Out[92]:
Latitude Longitude Year
0 41.739980 -87.555120 2008
1 41.726922 -87.599747 2018
2 41.994138 -87.734959 2007
3 41.881217 -87.730590 2018
4 41.874174 -87.668082 2021
In [93]:
fig = px.scatter_mapbox(map_marks[map_marks.Year == 2016].sample(5000), lat="Latitude", lon="Longitude",
                        color_discrete_sequence=["fuchsia"], zoom=9, height=650)
fig.update_layout(mapbox_style="open-street-map")
fig.update_layout(margin={"r":0,"t":0,"l":0,"b":0})
fig.show()

Observation::

It Looks like there is crime everywhere in Chicago, in the centre, you can see more points but also is a crowded place. The map of Chicago indicates that crimes are distributed evenly throughout the city, with no discernible clustering or concentration in specific locations. This suggests that criminal activity is prevalent across various neighborhoods and districts in Chicago, rather than being confined to particular areas. Such widespread distribution of crime may indicate systemic issues related to crime prevention, law enforcement, socio-economic factors, or community dynamics that affect the entire city. Addressing these underlying factors requires comprehensive strategies that focus on enhancing community safety, improving law enforcement efforts, addressing socio-economic disparities, and promoting community engagement and collaboration. Additionally, it underscores the importance of data-driven approaches to identify high-risk areas and allocate resources effectively to combat crime across the city.

Count of Arrest and Non-Arrest for Each Crime Type¶

In [94]:
# Group by 'Crime Type' and 'Arrest' and count occurrences
crime_arrest_counts = c_crime_df.groupby(['Crime Type', 'Arrest']).size().unstack(fill_value=0)

# Plot horizontally stacked bar plot
fig = px.bar(crime_arrest_counts, orientation='h', barmode='stack')
fig.update_layout(title='Count of Arrest and Non-Arrest for Each Crime Type', xaxis_title='Count', yaxis_title='Crime Type')
fig.show()
In [95]:
# Combine the percentages into a DataFrame
percentage_df = pd.DataFrame({'Crime Type': arrest_percentage.index, 'Arrest Percentage': arrest_percentage.values, 'Non-Arrest Percentage': non_arrest_percentage.values})

# Plot the horizontal bar chart using Plotly
fig = px.bar(percentage_df, x=['Arrest Percentage', 'Non-Arrest Percentage'], y='Crime Type', barmode='group')
fig.update_layout(title="Percentage of Arrest and Non-Arrest for Each Crime Type", xaxis_title="Percentage", yaxis_title="Crime Type", legend_title="Type")
fig.show()

The “VIOLATION” crimes look strongly correlated to the arrested rate once the 'cop' must be present in the location to declare the violation.

  1. Summarize Insights: Summarize the key insights and findings from the analysis.
    • How has crime in Chicago changed across years? Was 2021 really the bloodiest year in two decades?
    • Are some types of crimes more likely to happen in specific locations or specific time of the day or specific day of the week, or specific month of the year than other types of crimes?
  1. [Q] How has crime in Chicago changed across years? Was 2021 really the bloodiest year in two decades?
  • [A] Crime rates in chicago has taken a downwards trend from 2001 till 2021. A downward trend in crime rates is generally considered a positive sign for a community or city. It indicate that various factors such as improved law enforcement strategies, community policing initiatives, economic development, social programs, and advancements in technology have been effective in reducing crime. The year 2021 is not the bloodiest year from our observation and also because the crime rate has taken a downward trend since 2001. The downward trend in crime rates from 2001 to 2022 is a promising observation that reflects potential successes in crime prevention efforts, community engagement, and policy interventions. However, ongoing monitoring, evaluation, and collaboration among stakeholders are crucial to sustain and further build upon these achievements in ensuring community safety and well-being.
  1. [Q] Are some types of crimes more likely to happen in specific locations or specific time of the day or specific day of the week, or specific month of the year than other types of crimes?
In [96]:
# Specific Locations
plt.figure(figsize=(14, 10))
sns.heatmap(c_crime_df.pivot_table(index='Primary Type', columns='Location Description', aggfunc='size', fill_value=0), cmap='Blues')
plt.title('Frequency of Crime Types by Location')
plt.xlabel('Location Description')
plt.ylabel('Crime Type')
plt.xticks(rotation=90)
plt.show()

# Specific Times of the Day
plt.figure(figsize=(14, 10))
sns.heatmap(c_crime_df.pivot_table(index='Primary Type', columns='Hour', aggfunc='size', fill_value=0), cmap='Blues')
plt.title('Frequency of Crime Types by Hour of the Day')
plt.xlabel('Hour')
plt.ylabel('Crime Type')
plt.show()

# Specific Days of the Week
plt.figure(figsize=(14, 10))
sns.heatmap(c_crime_df.pivot_table(index='Primary Type', columns='Day', aggfunc='size', fill_value=0), cmap='Blues')
plt.title('Frequency of Crime Types by Day of the Week')
plt.xlabel('Day of Week')
plt.ylabel('Crime Type')
plt.show()

# Specific Month of the year
plt.figure(figsize=(14, 10))
sns.heatmap(c_crime_df.pivot_table(index='Primary Type', columns='Month', aggfunc='size', fill_value=0), cmap='Blues')
plt.title('Frequency of Crime Types by Month')
plt.xlabel('Month')
plt.ylabel('Crime Type')
plt.show()
No description has been provided for this image
No description has been provided for this image
No description has been provided for this image
No description has been provided for this image

Observation:

The analysis revealed distinct patterns in the occurrence of different types of crimes across various dimensions such as location, time of the day, day of the week, and month of the year.

  1. Specific Locations: Certain types of crimes showed a propensity to occur more frequently in particular locations. For example, theft-related crimes are more prevalent in crowded areas such as commercial districts in our case,TAVERN while violent crimes like assaults occur more frequently in residential neighborhoods or areas with high foot traffic.

  2. Time of the Day: The analysis indicated variations in crime rates depending on the time of the day. Some types of crimes exhibited diurnal patterns, with peaks during certain hours. For instance, burglaries are more common during daytime hours when residences are more likely to be unoccupied, while robberies may occur more frequently during evening hours.

  3. Day of the Week: Certain types of crimes displayed fluctuations in occurrence based on the day of the week. For example, alcohol-related offenses or disturbances might increase on weekends when social activities are more prevalent, while property crimes like theft and burglaries might occur more frequently on weekdays when residential areas are less populated due to work.

  4. Month of the Year: The analysis also revealed seasonal variations in crime rates. Certain types of crimes may exhibit increased activity during specific months due to factors such as weather conditions, holidays, or cultural events. For instance, thefts might rise during holiday shopping seasons, while vandalism or disorderly conduct might increase during summer months when outdoor activities are more common.

Overall, the observation highlights the complex interplay of various factors influencing crime patterns, including environmental, social, and temporal factors. Understanding these patterns is crucial for law enforcement agencies, policymakers, and community stakeholders to develop targeted interventions and strategies aimed at preventing and reducing crime in specific locations and times. Additionally, it underscores the importance of data-driven approaches in identifying high-risk areas and implementing effective crime prevention measures.

In [ ]: